The integration of a large number of components on a single IC chip requires complex interconnects. Ideally, the interconnect structures should be fabricated with minimal signal delay and optimal packing density. Because of their increasing importance, the qualities of the interconnect structures drastically affect the reliability and performance of fabricated integrated circuits. Currently, the interconnect structures are increasingly defining the limits in performance and density of modern very-large scale integrated (VLSI) circuits.
Until recently, aluminum and its alloys have been widely used as conductive materials for electrical interconnections because of attractive features such as low electrical resistivity and strong adhesion to silicon dioxide (SiO2), which is typically used as an interlayer dielectric. Unfortunately, however, as VLSI dimensions reach into the deep-submicron regime, aluminum and its alloys become limiting factors in achieving superior performance. For example, with decreasing dimensions, the design rules become restricted by aluminum reliability concerns such as electromigration, which in turn increases the potential for open circuits or voids, stress-induced void formation, hillocks at relatively low temperatures, or humidity-induced corrosion.
For the above reasons, and in an attempt to improve the performance, reliability and density of the interconnects, the microelectronics industry has recently migrated towards alternative metals to aluminum and its alloys. As such, studies have been done on copper and copper oxide, mainly because copper has become a promising interconnect material for the next generation of integrated circuits because of high conductivity, extremely low resistivity and good resistance to electromigration. Unfortunately, copper diffuses rapidly through SiO2 or other interlayer dielectrics, such as polyimides and parylenes. Copper diffusion in the interconnect structure can destroy active devices, such as transistors or capacitors, formed in the IC substrate. In addition, the adhesion of copper to interlayer dielectrics, particularly to SiO2, is generally poor and metal adhesion to the underlying substrate materials must be excellent to form reliable interconnect structures. Further, copper oxidizes easily at low temperatures and has poor adhesion to substrates. Copper has also low reaction temperature with most salicides and requires a high temperature for patterning by reactive ion etching.
In an attempt to overcome these disadvantages posed by copper interconnects, efforts have been made mainly at refining the understanding of the oxidation mechanisms in copper employed in the interconnect structures. For example, W. A. Lanford studied ion implantation as an effective way to passivate copper films. Lanford, W. A. et al., Low-temperature passivation of copper by doping with Al or Mg, in THIN SOLID FILMS, 234–41 (1995). By analyzing the growth mechanism for copper, Lanford observed that the oxidation rate could be reduced by adding only a very small concentration of dopant, such as Al or Mg, to the copper.
Similarly, the corrosion resistance of boron (B) implanted copper, particularly its mechanism, has been studied by P. J. Ding et al. in Investigation of the mechanism responsible for the corrosion resistance of B implanted copper, B 85 NUCL. INSTRUM. METHODS PHYS. RES., 260–63 (1994). By investigating the oxidation of boron implanted copper and copper oxide (Cu2O), Ding found that the oxidation rate of Cu2O implanted with boron is as low as that of copper metal (Cu) implanted with boron.
Other methods for overcoming the copper interconnect disadvantages have yet involved scraping the copper layer to remove the copper oxide immediately before the interconnect is formed, or using a barrier layer to passify the copper surfaces. For example, U.S. Pat. No. 4,987,750 describes the use of titanium nitride (TiN), tungsten (W), tungsten nitride (WN), zirconium nitride (ZrN), titanium carbide (TiC), tungsten carbide (WC), tantalum (Ta), tantalum nitride (TaN), or titanium tungsten (TiW) as barrier layers for copper.
Similarly, U.S. Pat. No. 5,447,599 to Li et. al. discloses the use of TiN(O) as a barrier layer material for copper. The copper is initially coated with a layer of titanium and a copper-titanium alloy is formed by heating. Unreacted titanium is then removed and the alloy is transformed to TiN(O) by rapid thermal anneal in ammonia and oxygen.
Many of these materials, however, also form nonconductive oxides, or have poor electrical or thermal conductivity, or a high thermal expansion. Further, many of these barrier layers exhibit instability at temperatures higher than 500° C. Since a silicon substrate is generally subjected to subsequent steps during the IC fabrication, such as annealing or reflow processes which require temperatures higher than 500° C., there is a need for a barrier layer that is stable at the high temperatures required for subsequent substrate processes steps. Also, since copper is increasingly used in the electrical interconnection technology, it is desirable to further improve the processes in which copper oxidation is effectively prevented, while its metallization resistance is kept low even after the substrate has undergone subsequent processing steps.
While copper diffusion and oxidation remain highly significant, high density integration in the microelectronics industry also faces the problem of large amount of heat generated as a result of controlling a large quantity of current. Thus, it is imperative to radiate the large quantity of heat generated to prevent an unacceptable rise in temperature in the semiconductor substrate. For this reason, heat radiation substrates have been used in the semiconductor industry with relative success. Nevertheless, a major drawback of the thermal radiating substrates used in conventional power semiconductor devices is their extremely complicated construction. Recently, aluminum nitride (AlN) has attracted attention, mainly because it has superior dielectric strength (140 to 170 kV/cm) and good thermal conductivity (90 W/m.° C.), and efforts have been made at trying to join AlN substrate to a copper member, as disclosed, for example, in U.S. Pat. No. 4,611,745 to Nakahashi et. al.
Accordingly, there is a need for an improved copper interconnect structure of an integrated circuit that would not oxidize to form a nonconductive material. A copper interconnect structure with a suppressed oxide growth layer having good electrical conductivity, good thermal conductivity, and low thermal expansion is also needed, as well as a simple process for forming such copper interconnect structure.